Dmitriy Smirnov
Senior Compiler Engineer at Arm
Greater Cambridge Area United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolDmitriy Smirnov is a Senior Compiler Engineer with over 8 years in high-throughput bioinformatics and a two-decade foundation in compiler and systems work, currently advancing machine learning toolchains at Arm in Cambridge. He combines deep compiler and numerical expertise—demonstrated by LLVM and TVM contributions improving FFT accuracy, operator lowering, and TFLite frontend support—with hands-on systems engineering across C/C++, Java, Python, and PL/SQL. Dmitriy has designed high-performance data pipelines and memory planners, implemented simulated-annealing allocation and non-recursive refactors, and led architecture and technical teams through full software lifecycles. His background spans low-level optimizer development (Intel, SPARC research) to production bioinformatics pipelines at EMBL-EBI, giving him rare fluency from hardware-adjacent codegen to large-scale data processing. He’s equally comfortable writing CI and tooling around Docker/Jenkins as he is tuning numerical edge cases in ML compiler passes.
8 years of coding experience
12 years of employment as a software developer
Национальный Исследовательский Ядерный Университет "МИФИ"
Github Skills (23)
tosca10
llvm10
machine-learning10
lsa10
mlr10
togaf10
compiler-compiler10
tensorflow10
compiler-optimization10
compiler10
tflite10
numerical9
computation9
spirv9
python9
Programming languages (5)
Github contributions (5)
apache/tvm
Jul 2020 - Nov 2022
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Role in this project:
ML Engineer
ML EngineerContributions:239 reviews, 45 commits, 48 PRs in 2 years 4 months
Contributions summary:Dmitriy primarily contributed to the TFLite frontend within the TVM project, adding and improving support for various TFLite operations, including arg_min/max, reshape, maximum/minimum, and gather. Their work involved implementing parser support, adding unit tests, and addressing potential issues in the existing code. The user's contributions facilitated the integration and optimization of models for different hardware targets within the TVM ecosystem.
metalvulkancompilertensoropencl
llvm/llvm-project
Oct 2023 - Mar 2025
The LLVM Project is a collection of modular and reusable compiler and toolchain technologies.
Role in this project:Back-end Developer
Back-end DeveloperContributions:16 reviews, 14 PRs, 30 comments in 1 year 4 months
Contributions summary:Dmitriy primarily contributed to the LLVM project by modifying code related to the TOSA (Tensor Operator Set Architecture) and MLIR (Multi-Level Intermediate Representation) dialects. Their work included increasing the accuracy of FFT2D/RFFT2D calculations and updating the tosa.negate operator. Additionally, the user worked on math.powf lowering for SPIR-V, ensuring that negative bases with fractional exponents produced NaN results and fixing bias handling for Winograd. These contributions suggest a focus on compiler optimization and numerical computation within the LLVM ecosystem.
compilerstechnologiesclangsubmittoolchain
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.